About feature extractor

bytedance / RLFN

Winner of runtime track in NTIRE 2022 challenge on Efficient Super-Resolution

Apache License 2.0

157 stars 22 forks source link

Closed Legenhoor closed 1 year ago

Legenhoor commented 2 years ago

How is the structure of a feature extractor ？ Is there just one Conv k3s1-Tanh-Conv k3s1 ?

fyan111 commented 2 years ago

Before Conv k3s1-Tanh-Conv k3s1， you need normalize input like VGG[ (input - mean) / std ]

zhuzhu18 commented 1 year ago

what is the number of the input and output channel in Conv k3s1-Tanh-Conv k3s1?